Meeting August 13, 2025

[danae404]

August 6, 2025 9am PST/4pm UTC https://us04web.zoom.us/j/77767978616?pwd=sAe8ppkDU1jNG2qSeyjmzmZyPiDcQA.1

This meeting will be August 13 9am PST/4pm UTC https://appleinc.webex.com/meet/destra

[danae404]

Comment anything you want to add to the meeting agenda, including any tests that you worked on since the previous meeting

[danae404]

#7 : should ::cue(*) target the root WebVTT element?

[danae404]

#4 : should ::cue target the root WebVTT element or its inline descendent?

[danae404]

#8 : how do we test that cue moves out of the way of controls?

[danae404]

Here are (I think) trivial bug fixes: #6 #5 . I merged the PRs to webkit, but I wasn't sure if I should wait to export them to WPT before we discuss them.

[danae404]

Meeting Notes:

• Our spreadsheet is missing data for the parsing and api tests. These tests each have their own subtests. I will update the spreadsheet with subtest results.

• Some WebVTT WPT use ::cue(*) selector incorrectly #7 :
  The section of the spec that discusses ::cue(selector) seems to indicate that the root can be matched by the selector, but later in the spec, under Example 22, this seems to be contradicted. Our tests should match the normative behavior, so we should assume for now that the root can be matched, and that these tests are correct. However we should file an issue on the WebVTT spec about this inconsistency and have it fixed, then we can revise the tests if necessary. I will file this issue.

• Some wpt ref files place font styling on the inline element, affecting line height #4 :
  We agree that ::cue should target the root, block level element. I will post a pr on wpt for these test changes and wait for others to approve. I believe Firefox does not match this behavior, so Firefox folks will take a look to see if they need to make changes to their implementation once the wpt pr is approved.

• Tests with controls showing are arbitrary #8 :
  This issue has been discussed before here and here. A solution has not been decided upon yet. I will attend the next TTWG meeting tentatively on August 28 and bring it up. In the meantime, these tests should be deleted.

• line_50_percent.html: margin-top: -4.5px in ref seems unnecessary #6 :
  This test was likely written before the spec made the default line alignment "start" (equates to the top edge for horizontal cues). I will post a pr removing this margin to wpt.

• WebVTT rendering test seems to calculate incorrect x-position of cue in ref file #5 :
  Based on the spec, the default position alignment is center. The "right" or "left" value of the ref cue needs to take that into account. If the position setting is 80%, our ref cue needs to be placed so that its center is length of video * 0.8 pixels to the right of the video's left edge. I will check out if these tests calculate that value correctly and post a pr if necessary.

Lastly, we discussed procedures for discussing/committing test changes. When someone finds a bug in the test suite, they will file an issue in this repo (interop-webvtt). The issue will link to an issue/pr on the webvtt spec and/or wpt. We will discuss any open issues at the following meeting. Being able to discuss problems asynchronously on GitHub should hopefully allow us to iterate more quickly, while hopefully ensuring that people's viewpoints are heard before changes are committed. I have posted a few pr's to WebKit, but I will close these and copy them to wpt so they are more easily accessible.

Our next meeting will be August 27. Thanks everyone!